Transport property measurements of polymer electrolytes

PII:

Electrochimica Acta, Vol. 43, Nos 10±11, pp. 1387±1393, 1998 # 1998 Published by Elsevier Science Ltd. All rights reserved Printed in Great Britain 0013±4686/98 $19.00 + 0.00 S0013-4686(97)10069-X

Transport property measurements of polymer electrolytes Anders Ferry,a* Marca M. Doe€b and Lutgard C. DeJongheb a Department of Experimental Physics, UmeaÊ University, S-901 87, UmeaÊ, Sweden Materials Science Division, Lawrence Berkeley Laboratory, 1 Cyclotron Road, Berkeley, CA 94720, U.S.A.

b

(Received 16 September 1996; accepted 22 March 1997) AbstractÐStraightforward electrochemical methods for determining the transport properties (bulk ionic conductivity, salt di€usion coecient, and cation transference number) of polymer electrolytes are described herein. The new technique for measuring t0+ is based on concentrated solution theory, and requires no assumptions to be made concerning ideality of the solution. The experimental methods are described and results on representative polymer electrolyte systems are presented in this paper. Cation transference numbers are found to be salt-concentration dependent and considerably less than unity or even negative over a wide concentration range. This implies that the cationic current is mainly carried by complexed ions. The presence of associated ionic species in the present systems is con®rmed by Raman spectroscopic evidence. Finally, the practical e€ects of a negative t0+ on cell operation and implications for device design are discussed. # 1998 Published by Elsevier Science Ltd. All rights reserved Key words: polymer electrolyte, transport properties, transference number, Raman spectroscopy, di€usion.

INTRODUCTION Operation of electrochemical devices utilizing polymer electrolytes requires the passage of ionic current. Insight into the nature and mobility of the current-carrying species is, therefore, essential to the successful design and utilization of polymer electrolyte materials in such devices. Three transport properties are needed for the full description of a binary polymer±salt solvent system [1]; these can be chosen to be ionic conductivity, the salt di€usion coecient and the cationic transference number (t0+). By de®nition, a transference number is the net number of moles of an ion constituent that crosses a reference plane ®xed with respect to the solvent when one Faraday of current is passed through the ®xed plane in the absence of concentration gradients. Measurement of the last has either proven to be dif®cult experimentally (eg the Hittorf method) or requires assumptions to be made about the ideality of the polymer solutions [2]. We have recently developed simple electrochemical methods to determine the transport properties *Author to whom all correspondence should be addressed.

of non-ideal binary polymer±salt solutions, including t0+, based on concentrated solution theory [3]. The transport properties calculated from these measurements are based on a macroscopic model and are, therefore, independent of speciation (eg t0+ includes transference of cations in both the free and complexed states). There is no need for detailed knowledge concerning the formation of ion pairs, triplets or higher order aggregates to gain a thorough understanding about ionic transport in the polymer electrolyte relevant to device operation. Ionic conductivity is measured as a function of salt concentration at the temperature of interest, using the ac impedance method. Salt di€usion coef®cients are determined from the relaxation of symmetrical cells galvanostatically polarized for short times. Data from the di€usion coecient experiments, concentration cells and current interrupts are used to calculate t0+ as a function of salt concentration for the polymer system of interest. Finally, the complete set of data may be used to calculate the thermodynamic factor (variation of the activity coecient with salt concentration). While no speci®c information is provided regarding the types of species present in the polymer solution,

1387

1388

A. Ferry et al.

the degree of non-ideality of the system can be determined. In contrast, Raman spectroscopy can be used to observe microscopic speciation directly, [4, 5], and thus is complementary to the methods described herein. EXPERIMENTAL P(EO)NaX (P(EO) = poly(ethylene oxide), ÿ X = CF3SOÿ 3 or N(CF3SO2)2 ) polymer ®lms were prepared as described previously [3]. Poly(propylene oxide) (PPO, average MW 4000, Aldrich) was carefully freeze dried using repeated pump±thaw cycles; dried salt was dissolved directly into the polymer using a magnetic stirrer and small amounts of acetonitrile. Acetonitrile was removed from the solution by further drying under vacuum at 1008C for 100 h. The abbreviation O:M is used throughout the paper to indicate the ratio of [CH2CH2O] (PEO) or [CH2CH(CH3)O] (PPO) groups to moles of salt (M indicates either Li or Na). LiTFSI (TFSI = (bis)tri¯uoromethanesulfonate imide, N(CF3SO2)2) was a gift from 3 M Company. NaTFSI was prepared by an ion-exchange reaction of LiTFSI with Na0.44MnO2. LiTFSI was dissolved in acetonitrile and excess Na0.44MnO2 added. The solution was stirred for 24 h and then ®ltered. The process was then repeated. Acetonitrile was removed from the solution by rotary evaporation, and the salt dried thoroughly in air and under vacuum. No lithium was detected in the elemental analysis of the resultant product (University of California Micro-analytical Laboratory, Berkeley, CA). Sodium (Alfa Products) was puri®ed as described previously [3], and rolled into foils approximately 5 mm thick, prior to use in cells. Lithium foils 5 mm thick from Cyprus-Foote Mineral Company were used as provided. The densities of the PEO based polymer electrolytes were measured by placing weighed ®lms in 10 ml speci®c gravity bottles ®lled with cyclohexane. The weight of the bottle before and after displacement of the liquid then allows the volume of the electrolyte to be calculated. Thermal characterization of PEO electrolytes was carried out using a Perkin-Elmer DCS-7 di€erential scanning calorimeter, and by measuring ac conductivity as a function of temperature. Cells of con®guration Na±polymer electrolyte±Na or Li± polymer electrolyte + Celgard 2500-Li were assembled under an inert atmosphere and equilibrated at the desired operating temperature (858C for the former, 258C for the latter) for several hours prior to testing. A computer-controlled MacPile II (Bio-logic, Claix, France) or PAR 173 potentiostatgalvanostat (EG&G) was used for the di€usion coecient measurements and current interrupt experiments. For concentration cell measurements, two polymer ®lms of di€erent compositions were overlapped edgewise, and sodium electrodes placed

at the ends, to provide a di€usion pathway of several cm. This is to prevent rapid relaxation of the salt concentration gradient and ensures that an accurate potential di€erence is obtained. Use of a high input impedance electrometer (Keithley 642) prevents cell polarization during measurement. A cell holder with a removable separator was used for the PPO concentration cells; the potential di€erence was determined immediately upon removal of the device separating two polymer solutions of di€ering compositions. A Solartron SI 1286 electrochemical interface and 1254 four-channel frequency response analyser was used to measure conductivities of polymer electrolytes in cells with blocking electrodes. FT-Raman spectra with a wavenumber resolution of 2 cmÿ1 were recorded at 858C with the temperature stability in the sample estimated to be 20.38C. The spectra were recorded with a Bruker IFS 66 with a Raman module FRA 106 and a continuous Nd:YAG laser (1064 nm) using a 1808 back scattering geometry. During measurements, the sample cell was placed in an evacuated thermostat. RESULTS AND DISCUSSION The methods described herein are best applied to true binary polymer±salt or liquid systems since concentrated solution theory dictates that n(n-1)/2 independent transport properties are needed to describe a system of n components [1]. Thus, the presence of low molecular weight additives in gelled polymer systems requires determination of additional transport properties (eg additional transference numbers and pairwise interaction parameters). In practice, these may be extremely dicult to obtain. These complications arise from the microscopic interactions among polymer, solvent and salt. Devices utilizing liquid electrolytes (eg LiClO4 in propylene carbonate) commonly have a porous separator such as a polypropylene ®lm between electrodes. Although the solvent-saturated separator resembles a gelled or plasticized polymer electrolyte in certain respects, the pores in the polymeric separator are much larger than those in the gelled systems [6]. The separator is, therefore, essentially inert, and may be regarded simply as a barrier reducing the e€ective cell area. Appropriate corrections should be made for pore volume fraction and tortuosity, but otherwise, the system may be considered a true binary salt solution, and the methods described herein are applicable. It should be noted, however, that rapid di€usional processes in low viscosity solvents may, in practice, be dicult to follow experimentally. Ionic transport occurs predominantly in amorphous (molten) regions of the polymer±salt complexes [7]. Therefore, the presence of crystalline polymer±salt complexes in some solutions at or near the operating temperature is a complicating

Transport property of polymer electrolytes factor in determining di€usion coecients and t0+. This is of particular concern in the case of PEO based electrolytes, which are frequently partially or fully crystalline at room temperature [8]. In some cases, the electrolyte is two-phase at the operating temperature, consisting of a poorly conducting crystalline phase dispersed in a molten conducting phase with a di€erent stoichiometry than that of the original polymer±salt mixture [9]. Any attempt to correlate transport properties with the salt concentration of the bulk polymer electrolyte will then lead to erroneous results, although some useful information may still be obtained. For this reason, thermal characterization of the polymer±salt system prior to obtaining transport properties is desirable. We note that the thermal history of PEO based electrolytes may be of importance in these studies, due to the slow kinetics of recrystallization for high molecular weight systems [10]. Phase diagrams of some polymer±salt systems are available in the literature [9]. If possible, temperature and salt concentration ranges should be chosen so that twophase regions are avoided. Figure 1 shows conductivity as a function of temperature for P(EO)15NaTFSI, as well as di€erential scanning calorimetry data. The conductivity shows a changeover from Arrhenius behavior to the predicted Vogel-Tammann-Fulcher behavior [11] at about 508C, close to the melting transition observed in the DSC scan. A eutectic was observed at 468C for compositions near P(EO)12NaTFSI. No melting transitions above 688C were observed over the concentration range used in this study (O:M = 6±500). Prud'homme and coworkers [12] obtained similar results on the PEO±NaTFSI system, with slight di€erences ascribable to their use of polymers with

Fig. 1. Conductivity as a function of temperature for an electrolyte of composition P(EO)15NaTFSI. The di€erential scanning calorimetry data is included and shows good agreement with the conductivity measurements.

1389

lower molecular weights than in this work (MW0 5  106). While these results suggest that the operating temperature of sodium batteries may be substantially lowered from the currently used 858C [13], we chose to make measurements at this temperature to avoid any possible complications from the presence of two-phase mixtures in the saltrich region of the phase diagram. The ®rst transport property to be measured is the bulk ionic conductivity at the temperature of interest. This most conveniently done using ac impedance techniques and cells with either blocking or non-blocking electrodes. In the latter case, the interfacial impedance may be determined in addition to the bulk electrolyte conductivity. It is well known that the interfaces between polymer electrolytes and alkali metals are composed of either a corrosion layer (for sodium) [14] or a solid electrolyte interphase region (SEI) [15] for lithium. Ion transport phenomena in the interfacial regions can generally be observed separately from analogous processes in the polymer electrolyte and, therefore, do not interfere with the transport property measurements. A possible exception is for t0+ and will be discussed below. and Results for the P(EO)nNaCF3SO3 P(EO)nNaTFSI systems are shown in Fig. 2. P(EO)nNaTFSI is generally more conductive than P(EO)nNaCF3SO3 over the entire composition range, but both systems have values comparable to, or somewhat greater than the lithium analogs. A relatively ¯at maximum is seen at intermediate salt concentrations in both cases. At low salt concentrations, conductivity decreases due to low numbers of charge carriers, whereas at high salt concentrations, the polymer segmental mobility is restricted by large densities of ionic crosslinks that reduce the ionic mobility. PPO(4000) is fully amorphous at room temperature; however, We note that Prud'homme and coworkers [16] have interpreted glass transition anomalies in terms of a micro-phase

Fig. 2. Conductivities of P(EO)nNaCF3SO3 (w) and P(EO)nNaTFSI (Q) electrolytes at 858C as a function of salt concentration. The numbers on the graph correspond to n for each composition. Data on P(EO)nNaCF3SO3 is taken from [3].

1390

A. Ferry et al.

separation occurring in complexes of PPO±LiClO4. Still, in the present study we treat the PPO±LiTFSI system as a single phase binary electrolyte, since a microscopic phase separation is unlikely to in¯uence the measurement of transport properties such as transference numbers, unless the domain sizes are relatively large or exist for long times relative to the observed phenomena. For the P(PO)nLiTFSI system, conductivities of about 1  10ÿ5 S cm are seen for moderately concentrated solutions (n = 15±50) at 258C. The molar conductivity (ionic conductivity normalized to salt concentration) exhibits the characteristic concentration dependency generally found in PPO based electrolytes [17±19]; a rapid increase from a minimum in dilute solutions passing through a maximum at an O:M ratio of about 40:1. Notably, the maximum conductivity is approximately an order of magnitude higher compared to PPO(4000) complexed with LiCF3SO3 [17]. In order to obtain salt di€usion coecients, Ds, symmetrical cells with non-blocking electrodes are polarized galvanostatically for short periods of time. When the current is turned o€, the induced concentration pro®le in the cell is allowed to relax. At long times after the current interrupt, the following equation holds true [3]: lnDF ˆ

p2 Ds t ‡ A1 L2

…1†

Where DF is the measured cell potential, Ds is the salt di€usion coecient, L is the electrolyte thickness and A1 a constant. Thus, a plot of lnDF vs time is linear and the slope is proportional to Ds. For the PEO±NaCF3SO3 system [3] at 858C, Ds decreases over nearly an order of magnitude as the salt concentration increases, ranging from about 8  10ÿ8 cm2 s for dilute solutions to about 1.2  10 ÿ8 cm2 s for P(EO)8NaCF3SO3. A small local minimum is seen at P(EO)80NaCF3SO3 for which Ds is about 5  10ÿ8 cm2 s. For PEO±NaTFSI, there is less variation in the upper concentration range; for O:M = 6±60, Ds is about 5  10ÿ8 cm2 s, a local minimum is seen at P(EO)120NaTFSI (Ds=3  10ÿ8 cm2 s), and the di€usion coecient increases gradually to about 7.6  10ÿ8 cm2 s for very dilute solutions (P(EO)495NaTFSI). For P(PO)nLiTFSI electrolytes at 258C, Ds increases with increasing concentration from n = 300 (Ds 00.7  10ÿ9 cm2 s), passing through a maximum of about 2.8  10ÿ9 cm2 s for a composition of P(PO)70LiTFSI. The observed increase closely resembles the concentration behavior of the molar conductivity. A relatively large number of hydroxyl terminal groups (O:OH 035:1) are present; solvated lithium ions may be preferentially coordinated to such polar moieties at low salt concentrations [19± 21]. It is important to stress that the salt di€usion coecients measured by this method, by de®nition, do not di€erentiate among species; di€usion due to

all free ions, ion-pairs and aggregates are included in these values. The complexation reactions of ions in highly non-ideal polymer electrolytes may involve multiple concurrent equilibria yielding species that vary greatly in their di€usivity; therefore, no simple relationships between salt di€usion coecients and composition are to be expected. Rather, the variation between salt di€usion coecient and concentration will be highly system speci®c and should be determined separately for each electrolyte and set of conditions. The variation of Ds with polymer composition is useful for determining the e€ect of concentration gradients upon the operation of working devices; large variations in Ds (and conductivity) with composition may promote premature failure of cells due to insucient ionic mobility in part of the electrolyte layer, or composite electrode. As with Ds, t0+ is a macroscopic quantity in this treatment and includes transference due to all cation-carrying species; ie ion-pairs, triplets and higher order aggregates. For the purpose of evaluating device performance, it is not necessary to know the details of speciation, although these may be of considerable interest to the polymer electrolyte scientist. It should also be noted that t0+ is de®ned in the absence of concentration gradients; although small gradients must be induced in experimental cells to make measurements, care should be taken to minimize perturbation of the system. To calculate transference numbers, the results from three separate experiments are combined, according to equation (2) [3]: mcF…pDs †1=2 dU t0ÿ ˆ 1 ÿ t0‡ …2† d ln c 4 Ds is the salt di€usion coecient from the restricted di€usion experiment, c is the molar salt concentration, dU/dlnc is the reciprocal of the local slope of concentration cell plots, and m is a slope derived from current interrupt data. The means for obtaining the latter two values are described below. Concentration cells of the form M±[electrolyte]n± [electrolyte]m±M are constructed, where M represents an alkali metal electrode and [electrolyte]n or m represents the polymer solution with salt concentration, n or m. Most conveniently, m is held constant, and n is varied; c refers to the concentration of [electrolyte]n and the potential reading is taken to be positive for n
F dU 2RT d ln C

…3†

These plots, however, deviate markedly from linearity, requiring knowledge of the thermodynamic fac-

Transport property of polymer electrolytes tor (variation of activity with concentration). A second experiment is necessary to obtain sucient information for the transference number calculation. For this, symmetrical cells are assembled and galvanostatically polarized by applying a current i for a time ti. Again, conditions are chosen so that there is minimal perturbation, sucient to establish concentration gradients at the electrodes, but not at the center of the cell. A potential reading taken shortly after the current is interrupted can then be correlated with the data from the concentration cell. Thus, the amount of salt in one segment of the polymer ®lm can be directly determined, as in the Hittorf method, without the need for sectioning and weighing the electrolyte ®lm. In equation (2), m is the initial slope of the cell potential vs it1/2 from i several experiments in which i and/or ti is varied. The potential reading DF should be taken before signi®cant salt di€usion occurs, but after double layer charging has dissipated. The latter is usually rapid (on the order of milliseconds) and the former slow (minutes to hours for polymer electrolytes). By plotting DF vs the dimensionless time T: Tˆ

t1=2 i

t1=2 ‡ …t ÿ ti †1=2

…4†

the potential di€erence corresponding to the concentration pro®le at the time of current interrupt can be retained by linear extrapolation back to T = 1 [23]. Figure 3 shows an example of this calculation. Obviously, it may be dicult to determine t0+ accurately if the overpotential in the cell is dominated by the sei or corrosion layer, and relaxation processes are on similar time scales as that of the polymer electrolyte. Some information regarding this may be obtained from ac impedance measurements on cells with non-blocking electrodes. The usual precautions with regard to purity and dryness should, of course, be taken prior to cell assembly. The systems used in this study were generally well behaved, and no unusual diculties were encountered. If high interfacial impedance is a persistent problem, less strongly reducing electrode materials may be used in place of metallic lithium or sodium, eg two phase Li or Na alloys [24±26]. Care should be taken to avoid changing the composition of these electrodes, as will occur if a substantial current is passed for a long time. Cationic transference numbers for all of the polymer systems in this study were strongly composition dependent. For the P(EO)nNaCF3SO3 system, t0+ values close to 0 were found for n = 40±120; for compositions more saltrich that P(EO)40NaCF3SO3 t0+ becomes negative [3]. For the PEO±NaTFSI system t0+ is negative over the entire range investigated, O:M = 120±7:1, decreasing slightly with increasing concentration from a value of t0+=ÿ0.27 for O:M=120:1 and reaching a low of t0+= ÿ 1.15 for

1391

Fig. 3. Plot of DF vs dimensionless time T (equation (4)) for a typical current interrupt experiment (composition P(EO)7NaTFSI; i = 0.073 mA/cm2, ti=30.35 s). The dashed line shows a linear extrapolation back to T = 1, corresponding to the time of current interrupt.

P(EO)7NaTFSI in the most concentrated solution studied. A more complex relationship between t0+ and salt concentration is seen for the PPO-LiTFSI system. The lithium ion transference number is negative for compositions O:M = 30±200, nearing 0 for the most dilute solutions, but reaches a minimum of ÿ2.3 at O:M = 70. Interestingly, this matches the point at which the salt di€usion coecient is at a maximum and roughly corresponds to the observed maximum in molar conductivity. The consistent observation of negative t0+ values for polymer electrolytes in all but the most dilute solutions may, at ®rst glance, seem surprising. There is, however, ample precedence for negative transference numbers in aqueous electrolytes [27]. This is best understood in terms of the deviation of the solution from ideality caused by ion complexation reactions. This may be further quanti®ed by relating the transference numbers to the thermodynamic factor [3]. Since t0+ is a macroscopic quantity, all cations, whether free or complexed, are included. Thus, a negative t0+ indicates that solvated ``free'' cations are relatively immobilized by the host matrix and that long range cationic transport predominantly involves correlated motions of cations and anions in the form of negatively charged aggregates of ions. In concentrated solutions, like those typically used in devices, the separation between ions may be only about 10 AÊ, and interchange among sites (ion pairs, triplets, aggregates, ions complexed by polymer, etc) occurs rapidly. Thus, to interpret ion transport processes in terms of discrete species is undoubtedly an oversimpli®cation. However, the notion that distinct associations of ions exist, at least over short time scales, is supported by extensive spectroscopic evidence [4, 5, 28]. For instance, in Fig. 4, the multicomponent nature of the Raman active ns(SO3) mode reveals that the anions experience several di€erent potential energy environments in the

1392

A. Ferry et al. bility is indeed seen for Na±P(EO)20NaCF3SO3±P2 NaxCoO2 cells than for Na±P(EO)8NaCF3SO3±P2 NaxCoO2 analogs [13]. CONCLUSIONS

Fig. 4. Raman spectra, ns(SO3) region, of P(EO)nNaCF3SO3 electrolytes at 858C. Components at 1033, 1037 and 1058 cmÿ1 are indicated by dotted lines. (Assignments in the text are taken from [28].)

P(EO)n NaCF3SO3 solutions. Components at 1033, 1037 and 1047 cmÿ1 have been attributed to ``free'' anions, NaCF3SO3 pairs and negatively charged triple ions, and positively charged Na2CF3SO3 triplets respectively [28]. The feature at around 1058 cmÿ1 has been assigned to larger aggregations of ions [28], possibly precursors to salt precipitation, and is only observed at higher concentrations in the present study (O:M < 55:1). We note that the existence of negatively charged ionic aggregates has been inferred from spectroscopic data in an endo-methylated PPO(4000)±LiCF3SO3 system by one of us [19]. It has been pointed out that a lithium battery can operate successfully even when t0+=0 [29], provided that ion pairs are mobile and anions can migrate in a direction opposite that of the ion pair di€usion. A similar logic applies when t0+ is negative. The lower the cationic transference number, however, the stronger the tendency towards development of concentration gradients in the cell. The steady-state concentration gradient during discharge is given by equation (5) [30]: Dc I…1 ÿ t0‡ † 1 L FDs

…5†

where L is the thickness of the electrolyte. If local regions of the electrolyte have compositions that exhibit poor transport properties, premature failure will occur. Computer modeling of Na± P(EO)12NaCF3SO3±Na cells show that a steeper concentration pro®le is generated at the anode than at the cathode [31] upon passage of current. Because the transference number and salt di€usion coecient decrease with increasing salt concentration for this system, precipitation at the anode will occur before a limiting current at the cathode is reached. Although it is a common practice to use highly concentrated electrolytes to prevent salt depletion, these results indicate that better cell performance will be obtained with less highly concentrated solutions. Markedly better rate capa-

Electrochemical techniques for determining the three transport properties necessary for describing the behavior of a single phase, binary polymer±salt electrolyte are discussed. These are 1) ionic conductivity, 2) salt di€usion coecient, and 3) the cationic transference number, t0+. A new, theoretically rigorous method for determining t0+ is presented, which has the advantage of experimental ease. Possible complications arising from the presence of solid electrolyte inter-phases, crystalline phases in the polymer±salt mix and liquid additives such as those used in gelled electrolytes are noted. It is found that the cationic transference number varies with the salt concentration in the polymer, and is often negative, especially for concentrated solutions. This is a direct result of the marked non-ideality of the polymer electrolytes, and indicates that cationic transport takes place primarily in the form of migration and di€usion of complexed ions. As a consequence, large concentration gradients may develop in working cells, leading to premature failure. For instance, mathematical modeling [31] of devices containing concentrated PEO±NaCF3SO3 electrolytes, shows that salt precipitation will ocur at the anode, and better rate capability is indeed obtained when less concentrated electrolyte solutions are used [13]. ACKNOWLEDGEMENTS This work was supported by the Assistant Secretary for Energy Eciency and Renewable Energy, Oce of Transportation Technologies, Oce of Advanced Automotive Technologies of the U.S. Department of Energy under Contract No. DE-AC0376SF00098, A. Ferry would like to than Stiftelsen Blance¯or-Boncompagni and JC Kempes Minnes Stipendiefond for ®nancial support. We would also like to thank Dr. Marc Doyle of DuPont for many helpful discussions. REFERENCES 1. J. Newman, Electrochemical Systems, Prentice-Hall, Englewood Cli€s, NJ (1991). 2. M. Doyle and J. Newman, J. Electrochem. Soc. 142, 3465 (1995). 3. Y. Ma, M. Doyle, T. Fuller, M. M. Doe€, L. C. DeJonghe and J. Newman, J. Electrochem. Soc. 142, 1859 (1995). 4. S. Schantz, L. M. Torell and J. R. Stevens, J. Chem. Phys. 94, 6862 (1991). 5. W. Huang, R. Frech and R. A. Wheeler, J. Phys. Chem. 98, 100 (1994).

Transport property of polymer electrolytes 6. M. Doyle, J. Newman, A. S. Godz, C. N. Schmutz and J-M. Tarascon, J. Electrochem. Soc. 143, 1890 (1996). 7. C. Berthier, W. Gorecki, M. Minier, M. B. Armand, J. M. Chabagno and P. Rigaud, Solid State Ionics 11, 91 (1983). 8. P. V. Wright, in Polymer Electrolyte Reviews-1, (Edited by J. R. MacCallum and C. A. Vincent), 61, Elsevier Applied Science, London (1987). 9. D. Fauteux, in Polymer Electrolyte Reviews-1, (Edited by J. R. MacCallum and C. A. Vincent), 121, Elsevier Applied Science, London (1987). 10. P. Ferloni, G. Chiodelli, A. Magistris and M. Sanesi, Solid State Ionics 18(19), 265 (1986). 11. M. A. Ratner, Mat. Forum 15, 1 (1991). 12. M. Perrier, S. Besner, C. Paquette, A. Vallee, S. Lascaud and J. Prud'homme, Electrochim. Acta 40, 2123 (1995). 13. M. Doe€, S. J. Visco, Y. Ma, M. Peng, L. Ding and L. C. DeJonghe, Electrochim. Acta 40, 2205 (1995). 14. K. West, B. Zachau-Christiansen, T. Jacobsen, E. Hiort-Lorenzen and S. Skaarup, Brit. Polymer J. 20, 243 (1988). 15. D. Fauteux, J. Electrochem. Soc. 135, 2231 (1988). 16. C. Vachon, M. Vasco, M. Perrier and J. Prud'homme, Macromolecules 26, 4023 (1993).

1393

17. I. Albinsson, B-E. Mellander and J. R. Stevens, J. Chem. Phys. 96, 681 (1992). 18. I. Albinsson, B-E. Mellander and J. R. Stevens, Solid State Ionics 72, 177 (1994). 19. A Ferry, J. Phys. Chem. B, 101, 150 (1997). 20. J. Manning, R. Frech and W. Huang, Polymer 31, 2245 (1990). 21. A. Bernson and J. Lindgren, Polymer 35, 4842 (1994). 22. A. Bouridah, F. Dalard, D. Deroo and M. Armand, J. Applied Electrochem. 17, 625 (1987). 23. J. Newman, personal communication. 24. J. O. Besenhard, M. Hess and P. Komenda, Solid State Ionics 40(41), 525 (1990). 25. J. O. Besenhard, P. Komenda, A. Paxinos and E. Wudy, Solid State Ionics 18(19), 823 (1986). 26. B. A. Boukamp, G. C. Lesh and R. A. Huggins, J. Electrochem. Soc. 128, 725 (1981). 27. RA Robinson, RH Stokes. 1959. Electrolyte Solutions. 2nd edn. Butterworths, London. p. 161. 28. P. Jaconsson, I. Albinsson, B-E. Mellander and J. R. Stevens, Polymer 33, 2778 (1992). 29. G. G. Cameron, J. L. Harvie and M. D. Ingram, Solid State Ionics 34, 65 (1989). 30. M. Doyle, T. F. Fuller and J. Newman, Electrochim. Acta 39, 2073 (1994). 31. Y. Ma, Ph.D Thesis, University of California at Berkeley, 1996.